AxxonSoft announces the launch of the 4.4 version of Axxon Next intelligent Video Management System

AxxonSoft has launched version 4.4 of the Axxon Next intelligent VMS. This version includes new functions of neural analytics and camera management, enhanced GUI and Web Client, and many other enhancements and improvements. Supported body temperature measurement with a number of thermographic camera models. Measurement results appear as captions over video and are saved to an archive.

Upon discovering an individual with abnormal body temperature, the camera creates an alarm event, and sends it to Axxon Next. The user can set up an auto response scenario to notify operators or responsible staff, start video recording, etc. Based on alarm events, the user can quickly find videos of individuals with elevated body temperature.

• Mask detection - Based on neural network algorithm for facial recognition, this tool detects the presence or absence of a mask on a face.
• Social distancing violation detection - Based on Behavior Analytics, close-standing people detection helps in social distancing enforcement. Along with non-contact body temperature measurement and mask detection, this function is included in a dedicated anti-pandemic solution.

Video analytics

The neural network analyses video and generates data on the postures of people in the FOV

AI analytics - Neural tracker recognises and tracks moving objects of a specified type, e.g., individuals or vehicles. The user can further apply Scene Analytics to recognised objects to detect their motion, stopping or lingering in an area, crossing a line, etc. Detection based on the neural tracker can be applied to complex scenes with a large amount of non-relevant detail, whereas classic motion detection would be drowned out by numerous false alarms.

In a specified time interval, the neural tracker counts objects of a specified type within a pre-defined area, and generates an alarm event upon reaching/exceeding a specified limit.

Posture detection

The neural network analyses video and generates data on the postures of people in the FOV. This data is processed by analytical algorithms which are capable of detecting specified postures, such as crouching, man-down, shooting, or raised arms.

Posture detection helps recognise potentially dangerous scenarios, such as:

• an individual crouched down next to an ATM could be a burglar
• an individual(s) in a shooting position and other(s) with raised arms - could be an armed robbery

Handrail holding detection helps in labour safety enforcement at production facilities, construction sites, working at height, etc. Posture detection–based counter tracks the total number of individuals within a specified area, and notifies staff upon reaching/exceeding the pre-defined limit.

Update required analytics

• Network hardware acceleration - Added support for neural network acceleration in NVIDIA and Intel GPUs, including Mustang-V100-MX8 (HDDL), Neural Compute Stick 2 and Intel HD Graphics
• Water level detection - This tool reads water level values from measurement scale video. Its output is represented in the Camera window with a colour-coded level indicator and, as an option, a numerical value. The user can use water level detection to monitor levels of any liquid in any basin or container.
• Facial recognition - Facial recognition now operates in real-time. Use the AxxonNet cloud service to create lists of Facial Templates and synchronise them across all Servers connected to the user’s AxxonNet account. Upon detecting a face, a list-specific response scenario is launched. The user can use this option to create lists of VIP and/or unwanted visitors, and automatically notify retail and security staff on their entry.
• DetectorPack - Axxon Next detection tools are now grouped by type (core, AI, facial, LPR) and compiled into a separate DetectorPack module, subject to independent continuous development and delivery. This makes it possible to download and update required analytics between point releases of VMS. Axxon Next 4.4 includes DetectorPack 3.4.0.

Web client

The new Web Client functions include:

• Simultaneous search on multiple camera channels for specified faces, vehicle numbers, detection events, or time interval
• Criteria- (MomentQuest) and time-based (TimeSlice) search
• Building a motion heat map
• Alarms panel displays all active alarms across the Axxon domain
• H.265 playback is supported (in the Edge browser with hardware acceleration set to on)
• The user can select one of the two H.264 display modes: all frames, or I-frames only
• Select the default layout to be displayed after the Web Client launch
• The user can now group cameras, create camera lists and sort cameras within a list by their names or IDs

Camera management

• Recording on motion from embedded VMD - When the user adds a camera to their system, they can now quickly set up motion-based recording from on-board Video Motion Detection. For each added camera, the system automatically creates a VMD tool and a rule for automated recording to the specified file. When VMD triggering ends, video recording stops.
• Adding links to other cameras - Users can now include links to other cameras in a camera window. Clicking a link brings the user to the linked camera. This function facilitates object tracking between different camera FOVs.

User Interface

• Videowall management - A video wall is a set of display monitors physically and logically connected to act as a single screen. A video wall may include any monitor connected to any Client within the Axxon domain. In Axxon Next 4.4, the users can set up a video wall via a WYSIWYG GUI: monitor images on screen now match their physical layout.

User rights now include an option to manage other Clients' monitors, not only local ones. On each monitor, users can set up layouts and quickly add cameras by dragging their icons from the Objects panel (Devices tree) or interactive Map. Hot keys for video walls control were added as well.

• Geo Map management - The user can specify coordinates of a camera: latitude, longitude, and bearing. When they add a camera to a geo map, its icon appears in a location that corresponds to its coordinates. On the map, the user can now search geo objects by their names. These functions will be useful for large-scale and distributed systems, including Safe City projects.
• Temporary layout in archive mode - The user can now create a temporary layout that includes cameras selected for Archive (video footage) viewing. This is a convenient tool for simultaneous viewing of multiple camera feeds for event analysis. When another layout is selected, the temporary layout is automatically deleted.
• Fast access to detection triggering events - The users can now quickly access detection triggering events on any layout with the newly introduced events panel to the right of the camera window. The panel contains the list of most recent detection events from tools created for this particular camera.